CN112001531B - Wind power short-term operation capacity credibility assessment method based on effective load capacity - Google Patents

Abstract

The invention discloses a method for assessing the credibility of wind power short-term operating capacity based on payload capacity, which includes: obtaining wind speed data, using ARMA to predict wind speed in the next period, and generating the predicted wind power output time series; obtaining load data and inputting The results of the conventional unit combination in each period; only conventional units are considered to calculate the system load loss probability LOLP; after adding wind turbines, the system load loss probability LOLP is solved again; the short-term operating capacity credibility of wind power SOCC is iteratively solved through the hybrid search algorithm. This case proposes a definition of short-term operating capacity reliability of wind power based on payload capacity. The proposed evaluation method takes into account the real-time operating status of the system and reflects the impact of changes in real-time operating conditions such as wind power output and load usage on reliability. The proposed evaluation The method fully considers the real-time operating status of the system, and the short-term trustworthy capacity of wind power in different periods can reflect the contribution of wind power output to system reliability.

Description

Wind power short-term operating capacity credibility assessment method based on payload capacity

Technical field

The invention relates to a short-term operation capacity credibility evaluation method of wind power based on payload capacity, and belongs to the wind power capacity credibility evaluation technology.

Background technique

The large-scale access to intermittent energy represented by wind power around the world has brought about major changes in the way the power system operates. The introduction of uncertain and intermittent wind power on the power generation side brings great challenges to the power system in maintaining a balance between power generation and load demand. Since the output of wind turbines is intermittent, the load capacity of wind turbines with the same capacity is different from that of conventional units. Therefore, in the power system adequacy measurement analysis, the power department treats wind turbines differently from conventional units.

Existing research on the credibility of wind power capacity mainly focuses on the planning state and is not applicable to the field of power dispatching. The reliability of planned wind power capacity cannot be directly applied to dispatching operations, otherwise it may lead to improper load distribution and excessive backup arrangements, which is contrary to the requirements of economic operation and leads to serious wind curtailment. The current short-term operating capacity reliability assessment method of wind power ignores the process of iterative solution of reliability indicators, so the assessment method lacks accuracy.

Contents of the invention

Purpose of the invention: In order to overcome the deficiencies in the existing technology, the present invention provides a short-term operation capacity credibility assessment method of wind power based on payload capacity. Based on the payload capacity, the short-term operation capacity credibility of wind power is proposed. The definition of SOCC; at the same time, based on the existing wind power forecasting method, ARMA is used to predict wind speed and generate wind power output time series; the ACD method is used to solve the system load loss probability LOLP before and after adding wind turbines; the short-term operating capacity of wind power is iteratively solved through a hybrid search algorithm Credibility SOCC.

Technical solution: In order to achieve the above objects, the technical solution adopted by the present invention is:

A method for assessing the credibility of wind power short-term operating capacity based on payload capacity, including the following steps:

S1. Obtain wind speed data and use the ARMA model to predict the wind speed time series V _t ;

S2. Convert the wind speed time series V _t into the wind power output time series P _wt ;

S3. Obtain load data and thermal power unit combination results for each period;

S4. Only consider the initial system composed of K thermal power units, and use the ACD method to calculate the initial system load loss probability LOLP _t,K ;

S5. Add M wind turbine units to the initial system, and use the ACD method to calculate the load loss probability LOLP _t,M+K of the new system;

S6. Use the hybrid search algorithm to iteratively solve the short-term operation credible capacity C _W (t) of wind power;

S7. Solve the short-term operating capacity credibility of wind power SOCC _t based on the short-term operating credible capacity of wind power C _W (t).

Preferably, in step S1, obtaining wind speed data and using the ARMA model to predict the wind speed time series V _t includes the following steps:

S11. Preprocess the wind speed data to obtain the time series {x _t }. First, calculate the autocorrelation function ρ _k of {x _t } and observe whether it quickly decays to near 0; if {x _t } does not meet the stationarity requirements, then It needs to be judged again after first-order difference processing; if the stationarity requirements are still not met after two differences, {x _t } cannot use ARMA to predict wind speed;

S12. Model identification and order determination, based on the autocorrelation coefficient ρ _k and partial autocorrelation coefficient of {x _t } Select an appropriate model and use the AIC criterion to select an appropriate model order;

The autocorrelation function (ACF) is defined as:

Among them: γ _k is the covariance, γ ₀ and σ _x ² are the variance, μ _x is the mean, k is the lag order (positive integer), n is the number of observations (the number of observations in the sequence);

The partial autocorrelation coefficient (PACF) is defined as:

The AIC guidelines are defined as:

Among them: n is the number of observations; p and q are the order of the model; σ _a is the variance of the residual when fitting the model; the order and parameters corresponding to the minimum AIC value are the optimal orders and parameters of the ARMA model ;

S13. Use maximum likelihood estimation to estimate model parameters, and call the Estimate function in MATLAB to implement the parameter estimation function;

S14. Use the DW statistic (Durbin-Watson Statistic) to test whether there is a first-order correlation in the model. If the output DW function value is close to 2, it is considered that there is no first-order correlation;

S15. To predict the wind speed time series V _t , the mathematical expression of the autoregressive moving average model ARMA is as follows:

x _t ＝a ₁ x _t-1 +a ₂ x _t-2 +···+a _p x _tp +ε _t -b ₁ ε _t-1 -b ₂ ε _t-2 -···-b _q ε _tq

t＝1,2,···,N

Among them: ε _t is the white noise sequence; a ₁ , a ₂ ,···,a _p is the autoregressive coefficient; b ₁ ,b ₂ ,···,b _q is the moving average coefficient; p and q are the orders of the model Number; {x _t } is the current time series value.

Preferably, in step S2, the relationship between the wind speed time series V _t and the wind power output time series P _wt is as follows:

Among them: V _ci , V _co , and V _r are the cut-in wind speed, cut-out wind speed, and rated wind speed of the wind turbine respectively; V _t is the wind speed at the current moment; A, B, and C are functions of V _ci and V _r ; P _r is The rated output of the wind turbine; the expressions of A, B, and C are as follows:

Preferably, in the step S3, the load data is represented by the probability pulse of the time series load; assuming that the research period is T hours, the load level at the tth hour is L _t , t=1,2,...,T, the load at the tth hour The probability of horizontal occurrence is (The purpose of this step is to simplify the solution of the reliability index in the ACD method), then the load data during the study period is expressed as L={L ₁ , L ₂ ,···,L _t ,···,L _T } .

Preferably, in step S4, for the initial system, the ACD method is used to calculate the initial system load loss probability LOLP _t,K , which includes the following steps:

S41. The effective capacity distribution of thermal power unit i is expressed as:

Among them: i=1,2,…,K, C _i is the installed capacity of thermal power unit i, is the effective capacity distribution of thermal power unit i, P _FOR,i is the random outage probability of thermal power unit i;

S42. The installed capacity C _i of thermal power unit i is a discrete random variable. The v-order moment of the discrete random variable C _i is defined as:

p _i =1-P _FOR,i

Among them: p _i is the normal operation probability of thermal power unit i, is the v-order moment of the installed capacity of thermal power unit i, α _v is the v-order moment of the effective capacity of thermal power unit, v is a positive integer;

S43. Convert each order moment into each order central moment:

Among them: M _v is the v-order central moment of the effective capacity of the thermal power unit, Take the permutation and combination of n for v;

S44. Calculate the cumulants of each order through the center distance of each order. The relationship between the first eight order cumulants and the central moments of each order is as follows:

Among them: K _v is the v-order accumulation of the effective capacity of the thermal power unit;

The equivalent effective capacity of each stage of S45 and K thermal power units is expressed as:

Among them: KS _v is the v-order cumulative amount of the equivalent effective capacity of K thermal power unit, K _{i, v} is the v-order cumulative amount of thermal power unit i;

S46. Use the Edgeworth series expansion to express the distribution of the equivalent effective capacity as a cumulative amount. The distribution function of the effective capacity of K thermal power units after loading is expressed as F _K (x):

Among them: F _K (x) is the probability that the power generation capacity provided by K thermal power units after loading is less than x, N (x) is the standard normal density function, N ^(γ) (x) is the γ order derivative of N (x) , g _v is the v-order normalized cumulant (the purpose of introducing this term is to simplify the series form), σ is the standard deviation;

S47. For the load level L _t at hour t, use F _K (L _t ) to express the probability that the power generation capacity at hour t is less than L _t , and obtain the load loss probability LOLP _t,K at hour t and F _K (L The relationship between _t ) is LOLP _t,K = F _K (L _t ).

Preferably, in step S5, M wind turbines are added to the initial system, and the ACD method is used to calculate the load loss probability LOLP _t,M+K of the new system, including the following steps:

S51. The effective capacity distribution of wind turbine j is expressed as:

Among them: j=1,2,…,M, P _W,j (t) is the output of wind turbine j at time t, is the effective capacity distribution of wind turbine j, P _{FOR, Wj} is the forced outage probability of wind turbine j;

S52. The output P _W, j of wind turbine unit j is a discrete random variable. The v-order moment of the discrete random variable P _W,j is defined as:

p _Wj =1-P _FOR,Wj

Among them: p _Wj is the normal operation probability of wind turbine j, P _W,j is the output of wind turbine j, β _v is the v order moment of the effective capacity of wind turbine, and v is a positive integer;

S53. Convert each order moment into each order central moment. The formula is as follows:

Among them: M _Wv is the v-order central moment of the effective capacity of the wind turbine, Take the permutation and combination of n for v;

S54. Calculate the cumulants of each order through the center distance of each order. The relationship between the first eight order cumulants and the central moments of each order is as follows:

Among them: K _Wv is the v-order cumulative amount of the effective capacity of the wind turbine;

S55, the equivalent effective capacity of each stage of the M wind power unit and the K thermal power unit is expressed as:

Among them: KW _v is the v-order cumulative amount of the equivalent effective capacity of M wind turbine generator units and K thermal power generator units, KS _v is the v-order cumulative amount of the equivalent effective capacity of K thermal power generator units, K _Wj,v is the wind turbine unit j The v-order cumulant;

S56. Use the Edgeworth series expansion to express the distribution of the equivalent effective capacity as a cumulative quantity. The distribution function of the effective capacity after loading of M wind turbine units and K thermal power units is expressed by F _M+K (x):

Among them: F _M+K (x) is the probability that the power generation capacity provided by M wind turbine units and K thermal power units after loading is less than x; N (x) is the standard normal density function; N ^(γ) (x) is N The γ-order derivative of (x); g _Wv is the v-order normalized cumulant (the purpose of introducing this term is to simplify the series form), σ is the standard deviation;

S57. For the load level L _t at the t-th hour, use F _M+K (L _t ) to express the probability that the power generation capacity at the t-th hour is less than L _t , and obtain the load-loss probability LOLP _t,M+K at the t-th hour and The relationship between F _M+K (L _t ) is LOLP _t,M+K =F _M+K (L _t ).

Preferably, in step S6, solving the short-term operation credible capacity C _W (t) of wind power includes the following steps:

S61. For the system composed of M wind turbine units and K thermal power units, calculate the maximum load loss probability of the system There is the following relationship with F _M+K (x):

in: is the maximum load loss probability of the system composed of M wind turbine units and K thermal power units, L _t is the load level at the t hour, and C _W is the total installed capacity of the wind turbine unit;

S62. For the system composed of M wind power units and K thermal power units, calculate the intermediate value of the maximum load loss probability of the system. There is the following relationship with F _M+K (x):

in: is the intermediate value of the maximum load loss probability of the system composed of M wind turbine units and K thermal power units, L _t is the load level at the t hour, and C _W is the total installed capacity of the wind turbine unit;

S63. Assign the load loss probability data of hour t:

R ₀ (t)=LOLP _t,K

R _W (t)=LOLP _t,M+K

S64. Solve R _q (t) iteratively by bringing in different ΔL _t ; let the number of iterations q=1, and compare the values of R ₀ (t) and R _mid (t): if the absolute value of the difference is greater than 5ε, that is |R ₀ (t)-R _mid (t)|≥5ε, then go to step S65; otherwise, go to step S66; where ε is the accuracy requirement of the iterative solution, ΔL is the wind turbine with a total installed capacity of C _W added to the initial system The amount of load the back can carry;

S65. Use the chord intercept method to solve the target value R _q (t). If |R _q (t)-R ₀ (t)|＞|R _q-1 (t)-R ₀ (t)|, it means the qth If an oscillation phenomenon occurs during the first iteration, proceed to step S66; otherwise, q=q+1, repeat step S65, and continue to use the chord-interception method to iteratively solve for the target value R _q (t);

S66. Use the dichotomy method to solve the target value R _q (t), where R _q (t) is the reliability index solved for the q-th iteration at the t-th hour, and enter step S67;

S67. If |R _q (t)-R ₀ (t)|≥ε, then q=q+1, repeat step S66, and continue to use the dichotomy method to iteratively solve the target value R _q (t); otherwise, enter step S68;

S68. Taking the load side as the entry point, and keeping the system reliability level unchanged before and after the wind turbine is connected, calculate the ratio of the load that can be carried to the installed wind power capacity:

R(C _g ,L)=R(C _g +C _W ,L+ΔL)

C _W (t)=ΔL _t

Among them: C _W (t) is the short-term reliable wind power operation capacity at the t hour; ΔL _t is the load that the initial system can carry at the t hour after adding wind turbines with a total installed capacity of C _W , and C _g is the initial The total installed capacity of the system (total installed capacity of thermal power units), R(C _g ,L) is the reliability level of the initial system when facing load L, R(C _g +C _W ,L+ΔL) is the reliability level of the initial system when facing load L The reliability level of the new system after adding wind turbines when facing load L+ΔL.

Preferably, in step S7, the wind power short-term operating capacity credibility SOCC _t is solved according to the following formula:

Among them: SOCC _t is the credibility of the short-term operating capacity of wind power at the t hour, C _W is the total installed capacity of the wind turbine unit, ΔL _t is the maximum capacity that the initial system can carry at the t hour after adding the wind turbine unit with a total installed capacity of C _W Load capacity.

Beneficial effects: The short-term operation capacity credibility evaluation method of wind power based on payload capacity provided by the present invention can conduct reliability evaluation on the premise of considering the real-time operating status of the system, and uses a hybrid search algorithm for the short-term operation credible capacity of wind power. Iterative solution can well avoid the impact of chord-intercept method search oscillation on the number of iterations, and can speed up the solution speed and accuracy; the present invention conducts reliability analysis based on the real-time operating status of the system, which can well reflect the timing characteristics of wind power output and its The impact of uncertainty on the credibility of wind power's short-term operating capacity; the short-term operating credible capacity of wind power in different periods can reflect the contribution of wind power output to system reliability.

Description of drawings

Figure 1 is an implementation flow chart of the present invention;

Figure 2 is a geometric interpretation diagram of trusted capacity based on payload capability in the present invention;

Figure 3 is a flow chart of the hybrid search algorithm of the present invention;

Figure 4 is a comparison chart of reliability indicators before and after wind power is connected;

Figure 5 is a comparison chart of the short-term operation credible capacity of SOCC and wind power in various periods.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings.

Figures 1 to 3 show a short-term wind power operating capacity credibility assessment method based on payload capacity, which includes the following steps:

S1. Obtain wind speed data and use the ARMA model to predict the wind speed time series V _t .

S11. When the ARMA model performs data prediction, the time series of the sample data must meet the requirements of stationarity; stationarity requires that the fitting curve obtained through the sample time series can still follow the existing form of "inertia" in the future. ” continues continuously, that is, the mean and variance are required not to change significantly; therefore, before conducting wind speed prediction modeling, it is necessary to detect whether the original wind speed sequence is a stationary random sequence; preprocess the wind speed data to obtain the time series {x _t }, and first calculate Autocorrelation function ρ _k of {x _t }, observe whether it quickly decays to near 0; if {x _t } does not meet the stationarity requirements, it needs to be judged again after first-order difference processing; if it still cannot be satisfied after two differences The stability requirement means that {x _t } cannot use ARMA to predict wind speed;

S12. Model identification and order determination, based on the autocorrelation coefficient ρ _k and partial autocorrelation coefficient of {x _t } Select an appropriate model; in the actual data processing and analysis process, the values of the orders p and q of the ARMA model are not unique. If they are estimated manually, errors will inevitably occur. Therefore, use the AIC criterion to select the appropriate model order. :

The autocorrelation function (ACF) is defined as:

Among them: γ _k is the covariance, γ ₀ and σ _x ² are the variance, μ _x is the mean, k is the lag order, and n is the number of observations;

The partial autocorrelation coefficient (PACF) is defined as:

Use the Akaike Information Criterion (AIC Criterion) to select the appropriate model order. This criterion uses the principle of maximizing the estimated value of the likelihood function to determine the model order; the AIC Criterion is defined as:

Among them: n is the number of observations; p and q are the order of the model; σ _a is the variance of the residual when fitting the model; the order and parameters corresponding to the minimum AIC value are the optimal orders and parameters of the ARMA model ;

S13. After determining the model structure and order and establishing the ARMA model, it is necessary to estimate the model parameters; commonly used parameter estimation methods include moment estimation, maximum likelihood estimation and least squares estimation. Moment estimation only uses p+q sample information, so there is too much information redundancy and poor estimation accuracy; in this case, maximum likelihood estimation is used to estimate model parameters, and the Estimate function is called in MATLAB to realize the parameter estimation function;

S14. Commonly used ARMA model testing methods include stationary reversibility test, overfitting test and residual analysis test. Some literature uses residual analysis test. This method tests whether the model residuals have a certain degree of randomness. If they do not have randomness, The selected model needs to be modified; in this case, the DW statistic (Durbin-Watson Statistic) is used to test whether there is a first-order correlation in the model. If the output DW function value is close to 2, it is considered that there is no first-order correlation;

S15. To predict the wind speed time series V _t , the mathematical expression of the autoregressive moving average model ARMA is as follows:

x _t ＝a ₁ x _t-1 +a ₂ x _t-2 +···+a _p x _tp +ε _t -b ₁ ε _t-1 -b ₂ ε _t-2 -···-b _q ε _tq

t＝1,2,···,N

Among them: ε _t is the white noise sequence; a ₁ , a ₂ ,···,a _p is the autoregressive coefficient; b ₁ ,b ₂ ,···,b _q is the moving average coefficient; p and q are the orders of the model Number; {x _t } is the current time series value.

S2. Convert the wind speed time series V _t into the wind power output time series P _wt .

After the wind speed V _t at time t is predicted according to the ARMA model, the wind power output power at time t can be calculated; using the above wind power prediction method, the uncertainty of the wind speed is included in the ARMA model, so for a certain time t, the wind speed V _t is relatively certain. The relationship between the wind speed time series V _t and the wind power output time series P _wt is as follows:

Among them: V _ci , V _co , and V _r are the cut-in wind speed, cut-out wind speed, and rated wind speed of the wind turbine respectively; V _t is the wind speed at the current moment; A, B, and C are functions of V _ci and V _r ; P _r is The rated output of the wind turbine; the expressions of A, B, and C are as follows:

S3. Obtain load data and thermal power unit combination results in each period.

The load data is represented by the probability pulse of the time series load; assuming that the research period is T hours, the load level at the t hour is L _t , t=1,2,...,T, and the probability of the load level occurring at the t hour is (The purpose of this step is to simplify the solution of the reliability index in the ACD method), then the load data during the study period is expressed as L={L ₁ , L ₂ ,···,L _t ,···,L _T } .

S4. Only consider the initial system composed of K thermal power units, and use the ACD method to calculate the initial system load loss probability LOLP _t,K .

S41. Thermal power units are commonly represented by a two-state model. The effective capacity distribution of thermal power unit i is expressed as Thermal power units are sorted according to power generation cost and loaded sequentially, then the effective capacity of the first K thermal power units/> with/> The relationship is as follows:

Among them: i=1,2,…,K, C _i is the installed capacity of thermal power unit i, is the effective capacity distribution of thermal power unit i, P _FOR,i is the random outage probability of thermal power unit i;

S42. As the scale of the power system expands, the amount of convolution and deconvolution operations each time is too large. Therefore, the ACD method uses cumulative quantities at each level to describe the load level of the system in each period and the random outage of the unit. This method makes the convolution The product operation is simplified to the addition operation of several cumulants, which greatly simplifies the calculation difficulty. After obtaining the cumulants of each order of the effective capacity distribution curve, the function value of each point on the curve can be obtained through the Edgeworth series expansion. Thus, the required reliability index is calculated; the installed capacity C _i of the thermal power unit i is a discrete random variable, and the v-order moment of the discrete random variable C _i is defined as:

p _i =1-P _FOR,i

Among them: p _i is the normal operation probability of thermal power unit i, is the v-order moment of the installed capacity of thermal power unit i, α _v is the v-order moment of the effective capacity of thermal power unit, v is a positive integer;

S43. The cumulative amount K _v is also a numerical characteristic of random variables. It can be obtained from moments of each order not higher than the corresponding order. To simplify the operation, convert each moment of order into central moments of each order:

Among them: M _v is the v-order central moment of the effective capacity of the thermal power unit, Take the permutation and combination of n for v;

S44. Calculate the cumulants of each order through the center distance of each order. The relationship between the first eight order cumulants and the central moments of each order is as follows:

Among them: K _v is the v-order accumulation of the effective capacity of the thermal power unit;

S45. It can be seen from the additivity of cumulants that when the effective capacity distributions of thermal power units are independent of each other, the distribution function of the equivalent effective capacity can be realized by the addition operation of the cumulants to replace the convolution operation, thus simplifying the calculation difficulty; therefore The equivalent effective capacity of K thermal power units at each level is expressed as:

Among them: KS _v is the v-order cumulative amount of the equivalent effective capacity of K thermal power unit, K _{i, v} is the v-order cumulative amount of thermal power unit i;

S46. Use the Edgeworth series expansion to express the distribution of the equivalent effective capacity as a cumulative amount. The distribution function of the effective capacity of K thermal power units after loading is expressed as F _K (x):

Among them: F _K (x) is the probability that the power generation capacity provided by K thermal power units after loading is less than x, N (x) is the standard normal density function, N ^(γ) (x) is the γ order derivative of N (x) , g _v is the v-order normalized cumulant (the purpose of introducing this term is to simplify the series form), σ is the standard deviation;

S47. For the load level L _t at hour t, use F _K (L _t ) to express the probability that the power generation capacity at hour t is less than L _t , and obtain the load loss probability LOLP _t,K at hour t and F _K (L The relationship between _t ) is LOLP _t,K = F _K (L _t ).

S5. Add M wind turbine units to the initial system, and use the ACD method to calculate the load loss probability LOLP _t,M+K of the new system.

S51. After adding the wind farm to the initial system, load the wind turbines first, and then load the conventional generating units in sequence; use a two-state model to describe the wind turbines, and the effective capacity distribution of wind turbine j is expressed as:

Among them: j=1,2,…,M, P _W,j (t) is the output of wind turbine j at time t, is the effective capacity distribution of wind turbine j, P _{FOR, Wj} is the forced outage probability of wind turbine j;

S52. The output P _W, j of wind turbine unit j is a discrete random variable. The v-order moment of the discrete random variable P _W,j is defined as:

p _Wj =1-P _FOR,Wj

Among them: p _Wj is the normal operation probability of wind turbine j, P _W,j is the output of wind turbine j, β _v is the v order moment of the effective capacity of wind turbine, and v is a positive integer;

S53. Convert each order moment into each order central moment. The formula is as follows:

Among them: M _Wv is the v-order central moment of the effective capacity of the wind turbine, Take the permutation and combination of n for v;

S54. Calculate the cumulants of each order through the center distance of each order. The relationship between the first eight order cumulants and the central moments of each order is as follows:

Among them: K _Wv is the v-order cumulative amount of the effective capacity of the wind turbine;

S55, the equivalent effective capacity of each stage of the M wind power unit and the K thermal power unit is expressed as:

Among them: KW _v is the v-order cumulative amount of the equivalent effective capacity of M wind turbine generator units and K thermal power generator units, KS _v is the v-order cumulative amount of the equivalent effective capacity of K thermal power generator units, K _Wj,v is the wind turbine unit j The v-order cumulant;

S56. Use the Edgeworth series expansion to express the distribution of the equivalent effective capacity as a cumulative quantity. The distribution function of the effective capacity after loading of M wind turbine units and K thermal power units is expressed by F _M+K (x):

Among them: F _M+K (x) is the probability that the power generation capacity provided by M wind turbine units and K thermal power units after loading is less than x; N (x) is the standard normal density function; N ^(γ) (x) is N The γ-order derivative of (x); g _Wv is the v-order normalized cumulant (the purpose of introducing this term is to simplify the series form), σ is the standard deviation;

S57. For the load level L _t at the t-th hour, use F _M+K (L _t ) to express the probability that the power generation capacity at the t-th hour is less than L _t , and obtain the load-loss probability LOLP _t,M+K at the t-th hour and The relationship between F _M+K (L _t ) is LOLP _t,M+K =F _M+K (L _t ).

S6. Use the hybrid search algorithm to iteratively solve the short-term operation credible capacity C _W (t) of wind power.

When wind turbines are not considered, the equivalent effective capacity distribution curve F _K (x) of K thermal power units is obtained through the ACD method; after adding M wind turbines, the equivalent effective capacity distribution curve F _M+K of the new system is obtained (x); When faced with the load L _t at hour t, the load loss probabilities are LOLP _t,K and LOLP _t,M+K respectively. Since the unit effective capacity distribution function F ( x) is more complicated. When the target value LOLP _t,K is known, the required x cannot be solved by the inverse function method, so iterative calculation can be performed by modifying x; solving LOLP _t,M+K ′=LOLP _t,K L _t + ΔL _t at this time, ΔL _t at this time is the short-term operation trustworthy capacity of wind power.

S61. Use the method of step S5 to calculate the maximum load loss probability of the system for the system composed of M wind power units and K thermal power units. According to the definition, assuming that only the system reliability index of conventional units is LOLP _t,K at hour t, and the system reliability index after adding the wind turbine unit with output P _Wt is LOLP _t,M+K ; if the new power supply is 100% If an ideal unit is 100% reliable, its reliability should be LOLP _St ^max . However, due to the random outage rate of the unit and other failure factors, the credible capacity of the new wind turbine unit is between 0 and the installed wind power capacity C _W ;/> There is the following relationship with F _M+K (x):

in: is the maximum load loss probability of the system composed of M wind turbine units and K thermal power units, L _t is the load level at the t hour, and C _W is the total installed capacity of the wind turbine unit;

S62. Use the method of step S5 to calculate the intermediate value of the maximum load loss probability of the system for the system composed of M wind power units and K thermal power units. There is the following relationship with F _M+K (x):

in: is the intermediate value of the maximum load loss probability of the system composed of M wind turbine units and K thermal power units, L _t is the load level at the t hour, and C _W is the total installed capacity of the wind turbine unit;

S63. When using the chord-interception method, oscillation will occur near the optimal value, causing the inability to converge to the optimal solution that meets the accuracy requirements, resulting in an increase in the number of iterations; when using the dichotomy method, compared to the chord-interception method The search speed is slow, but as the search scope continues to shrink, the optimal solution can always be found in the end; this case proposes a hybrid search algorithm based on the advantages and disadvantages of the comprehensive search planning state wind power capacity credibility algorithm; for the tth hour Loss of load probability data is assigned:

R ₀ (t)=LOLP _t,K

R _W (t)=LOLP _t,M+K

S64. Solve R _q (t) iteratively by bringing in different ΔL _t ; first determine the interval position of the target value, and use the bisection method to search accurately when it is close to the target value; if it is far away from the target value, use the chord intercept method to speed up the search Until it is close to the target value, switch to the dichotomy precise search; specifically: let the number of iterations q=1, compare the values of R ₀ (t) and R _mid (t): if the absolute value of the difference is greater than 5ε, that is |R ₀ (t)-R _mid (t)|≥5ε, then go to step S65; otherwise, go to step S66; where ε is the accuracy requirement of the iterative solution, ΔL is the wind turbine with a total installed capacity of C _W added to the initial system The amount of load the back can carry;

S65. Use the chord intercept method to solve the target value R _q (t). If |R _q (t)-R ₀ (t)|＞|R _q-1 (t)-R ₀ (t)|, it means the qth An oscillation phenomenon occurs during the first iteration, and R _q (t) swings near the target value. Therefore, it is necessary to use the dichotomy method for convergence iteration operation and enter step S66; otherwise, no oscillation occurs, q=q+1, repeat step S65, and continue. Use the chord-intercept method to iteratively solve for the target value R _q (t);

S66. Use the dichotomy method to solve the target value R _q (t), where R _q (t) is the reliability index solved for the q-th iteration at the t-th hour, and enter step S67;

S67. If |R _q (t)-R ₀ (t)|≥ε, it indicates that the calculation result meets the accuracy requirements, then q=q+1, repeat step S66, and continue to use the dichotomy method to iteratively solve the target value R _q (t ); otherwise, enter step S68;

S68. Taking the load side as the entry point, and keeping the system reliability level unchanged before and after the wind turbine is connected, calculate the ratio of the load that can be carried to the installed wind power capacity:

R(C _g ,L)=R(C _g +C _W ,L+ΔL)

C _W (t)=ΔL _t

Among them: C _W (t) is the short-term reliable wind power operation capacity at the t hour; ΔL _t is the load that the initial system can carry at the t hour after adding wind turbines with a total installed capacity of C _W , and C _g is the initial The total installed capacity of the system (total installed capacity of thermal power units), R(C _g ,L) is the reliability level of the initial system when facing load L, R(C _g +C _W ,L+ΔL) is the reliability level of the initial system when facing load L The reliability level of the new system after adding wind turbines when facing load L+ΔL.

S7. Solve the short-term operating capacity credibility of wind power SOCC _t based on the short-term operating credible capacity of wind power C _W (t).

Changes in real-time operating conditions such as wind power output and load utilization will lead to changes in reliability levels. Therefore, the short-term operating capacity credibility of wind power SOCC _t fully takes into account the real-time operating status of the system; the expression of short-term operating capacity credibility of wind power is:

Among them: SOCC _t is the credibility of the short-term operating capacity of wind power at the t hour, C _W is the total installed capacity of the wind turbine unit, ΔL _t is the maximum capacity that the initial system can carry at the t hour after adding the wind turbine unit with a total installed capacity of C _W Load capacity.

According to the above method, the final short-term operating capacity credibility of wind power can be obtained.

A specific embodiment of the short-term wind power operation capacity credibility assessment method based on payload capacity of the present invention is as follows:

Taking the IEEE-RTS79 reliability test system as an example for simulation, the total installed capacity of the system is 3105MW after removing the hydropower units in the system, and the nuclear power units are equivalent to conventional thermal power units. Using wind speed prediction data from a place in Jiangsu and modeling of wind power output, it is assumed that there are 108 wind turbine units with a total installed capacity of 162MW. The cut-in, rated and cut-out wind speeds of the units are 3.33, 13.55 and 22.22m/s respectively, and their random shutdown The luck rate is 0.04. The research period is 24 hours, and 400MW of reserve capacity is configured for each period.

Table 1 Wind power output forecast data

The data in the table is selected from the measured wind speed data of a wind farm in a certain place in Jiangsu. It is sampled every 10 minutes, with a total of 430 sampling points. The first 286 sampling point data are selected as original wind speed data, and the wind speed data of the next 144 sampling points are predicted. The wind power output prediction data in the table is obtained through the functional relationship between wind speed and wind turbine output, for a total of 24 hours.

Table 2 Load data

The data in the table is selected from the 23rd Sunday of the IEEE-RTS79 reliability test system. The above data are used to calculate the reliability index LOLP of the system before and after wind power is connected. The results are shown in Figure 4. Through comparison, it can be found that before the wind power is connected, the system reliability index LOLP changes with the fluctuation of the load, and the reliability in each period is different; after the wind power is connected, the system reliability is improved, but due to the randomness of the wind power output, As a result, the LOLP value fluctuates greatly in some periods. Although the wind power output at this time is a predicted value and there is a certain degree of error, it still proves that adding a certain amount of wind power output can help improve system reliability.

Table 3 SOCC and wind power short-term operation credible capacity in each period

The data in the table is obtained by using the hybrid search algorithm to iteratively solve the trustworthy capacity of short-term wind power operation based on the payload capacity-based definition of wind power short-term operation capacity proposed by the present invention. The calculation results of the short-term credible capacity and SOCC of wind power in each period are shown in Figure 5. The credible capacity of wind power in each period indicates the load that the new wind power output can carry when the system reliability is equal before and after wind power is connected. Through comparison, it can be found that due to the dual influence of the randomness of wind power in each period and maintaining the same reliability level, the short-term credible capacity of wind power is lower than the forecast output of wind power. By comparing Figure 4 and Figure 5, it can be found that when 14 o'clock and When the gap in system reliability levels before and after wind power is connected at 15:00 is large, the short-term operation trustworthy capacity and SOCC of wind power are both large; while when the gap in reliability levels between 1:00 and 24:00 is small, the short-term operation trustworthy capacity and SOCC of wind power are both large. Smaller, and the wind power utilization rate is above 95%.

Table 4 Comparison of trusted capacity search algorithms

The above SOCC solution examples are compared using the trusted capacity search algorithm proposed by the present invention and the chord intercept method and the bisection method. The results are shown in Table 4. Through comparison, it can be found that when the accuracy requirements are low, there is not much difference in the number of iterations required by the three methods; and when the accuracy requirements are high, the method in the present invention can still search for the optimal solution with the minimum number of iterations. Therefore, the advantage of the search algorithm proposed in this article is that it first judges the distance between the optimal solution and the iteration value, and then uses the chord-intercept method or the bisection method to search based on the distance judgment, which greatly avoids the chord-intercept method when the accuracy requirements are high. The resulting oscillation problem.

In summary, the present invention proposes a short-term operating capacity credibility definition of wind power that is suitable for considering the real-time operating status of the system; and proposes a reliability assessment based on the effective capacity distribution cumulant method, which can take into account the real-time operating status of the system. Reliability assessment is carried out under the conditions; a hybrid search algorithm is proposed for the iterative solution of the trustworthy capacity of short-term operation of wind power, which can well avoid the impact of chord-intercept method search oscillation on the number of iterations, and speed up the solution speed and accuracy; this paper The invention conducts reliability analysis based on the real-time operating status of the system, which can well reflect the impact of wind power output timing characteristics and uncertainty on the credibility of wind power short-term operating capacity; and the short-term operating credible capacity of wind power in different periods can reflect wind power output Contribution to system reliability.

The above are only the preferred embodiments of the present invention. It should be pointed out that those of ordinary skill in the art can make several improvements and modifications without departing from the principles of the present invention. These improvements and modifications can also be made. should be regarded as the protection scope of the present invention.

Claims (4)

1. A wind power short-term operation capacity credibility assessment method based on effective load capacity is characterized by comprising the following steps of: the method comprises the following steps:

s1, acquiring wind speed data, and predicting a wind speed time sequence V by using an ARMA model _t ；

S2, wind speed time sequence V _t Conversion to wind power output time sequence P _wt ；

S3, obtaining load data and a thermal power generating unit combination result of each period;

s4, only considering an initial system formed by K thermal power generating units, and calculating the load loss probability LOLP of the initial system by using an ACD method _t,K The method comprises the steps of carrying out a first treatment on the surface of the The method comprises the following steps:

s41, the effective capacity distribution of the thermal power generating unit i is expressed as:

wherein: i=1, 2, …, K, C _i Is the installed capacity of the thermal power generating unit i,is the effective capacity distribution of the thermal power unit i, P _FOR,i The probability of random outage of the thermal power unit i is obtained;

s42, installed capacity C of thermal power generating unit i _i For discrete random variables, a discrete random variable C is defined _i The v-order moment of (2) is:

p _i ＝1-P _FOR,i

wherein: p is p _i Is the normal operation probability of the thermal power generating unit i,is v-order moment of installed capacity of thermal power unit i, alpha _v V-order moment which is the effective capacity of the thermal power generating unit, wherein v is a positive integer;

s43, converting each order moment into each order center moment:

wherein: m is M _v Is the v-order central moment of the effective capacity of the thermal power generating unit,taking n as v and arranging and combining;

s44, obtaining the accumulation amount of each order through the center distance of each order, wherein the relation between the accumulation amount of the first 8 orders and the center moment of each order is as follows:

wherein: k (K) _v The v-order accumulated quantity is the effective capacity of the thermal power generating unit;

s45, the accumulated quantity of each step of equivalent effective capacity of the K thermal power generating units is expressed as follows:

wherein: KS (KS) _v The v-order cumulant for equivalent effective capacity of K thermal power generating units, K _i,v The v-order cumulative quantity of the thermal power unit i;

s46, the distribution of equivalent effective capacity is expressed by accumulation through edge worth series expansion, and F is used as a distribution function of the effective capacity after the loading of the K thermal power generating units _K (x) The representation is:

wherein: f (F) _K (x) The probability that the power generation capacity provided after loading of K thermal power generating units is smaller than x is that N (x) is a standard normal density function, N ^(γ) (x) Gamma derivative of N (x), g _v The normalized cumulative quantity is of v order, and sigma is the standard deviation;

s47, load level L for t hour _t Using F _K (L _t ) Indicating that the power generation capacity at t hour is less than L _t To obtain the probability of load loss LOLP at t hour _t,K And F is equal to _K (L _t ) Is related to LOLP _t,K ＝F _K (L _t )；

S5, adding M wind turbines into the initial system, and calculating new system load loss probability LOLP by using an ACD method _t,M+K ；

S6, iteratively solving wind power short-term operation credible capacity C through hybrid search algorithm _W (t); the method comprises the following steps:

s61, calculating maximum load loss probability of a system formed by M wind turbines and K thermal power plantsAnd F is equal to _M+K (x) The following relationship exists:

wherein:for the maximum load loss probability of a system formed by M wind turbines and K thermal power plants, L _t At load level at t hours, C _W The total capacity of the wind turbine generator is set;

s62, aiming at a system formed by M wind turbines and K thermal power plants, calculating the intermediate value of the maximum load loss probability of the systemAnd F is equal to _M+K (x) The following relationship exists:

wherein:is the intermediate value of the maximum load loss probability of a system formed by M wind turbines and K thermal power plants, L _t At load level at t hours, C _W The total capacity of the wind turbine generator is set;

s63, assigning the load loss probability data at the t hour:

R ₀ (t)＝LOLP _t,K

R _W (t)＝LOLP _t,M+K

s64 by bringing in different DeltaL _t Iterative solution R _q (t); let iteration number q=1, compare R ₀ (t) and R _mid Size of the value of (t): if the absolute value of the difference is greater than 5 ε, i.e. |R ₀ (t)-R _mid (t) | is not less than 5 epsilon, and the step S65 is carried out; otherwise, step S66 is entered; wherein epsilon is the precision requirement of iterative solution, deltaL is the total capacity of the initial system added loader is C _W The load quantity which can be borne by the wind turbine generator;

s65, solving target value R by chord cut method _q (t) if |R _q (t)-R ₀ (t)|＞|R _q-1 (t)-R ₀ (t) |, indicating that the oscillation phenomenon is generated during the q-th iteration, and proceeding to step S66; otherwise, q=q+1, repeating step S65, and continuing to iteratively solve the target value R using the chord cut method _q (t)；

S66, solving target value R by using dichotomy _q (t), wherein R is _q (t) obtaining a reliability index for the (t) th time and the (q) th time iteration, and entering step S67;

s67, if|R _q (t)-R ₀ (t) |ε or more, then q=q+1, repeat step S66, continue to iteratively solve the target value R using the dichotomy _q (t); otherwise, step S68 is entered;

s68, taking a load side as an access point, and calculating the ratio of the load quantity capable of being born more to the installed capacity of wind power under the condition that the reliability level of the system is kept unchanged before and after the wind power generation set is accessed:

R(C _g ,L)＝R(C _g +C _W ,L+ΔL)

C _W (t)＝ΔL _t

wherein: c (C) _W (t) is the wind power short-term operation credible capacity of the t hour; ΔL _t Adding installed total capacity to initial System of C _W The load quantity of multiple loads which can be born at t hours after the wind turbine generator, C _g R (C _g L) is the reliability level of the initial system in the face of load L, R (C) _g +C _W L+Δl) is the reliability level when a new system after adding a wind turbine into the initial system faces the load l+Δl;

s7, short-term operation credible capacity C based on wind power _W (t) solving the reliability SOCC of the short-term running capacity of wind power _t 。

2. The method for evaluating reliability of short-term running capacity of wind power based on payload capacity according to claim 1, characterized by: in the step S3, the load data is represented by probability pulses of time-series load; let the study period be T hours, and the load level at T hours be L _t T=1, 2, …, T, probability of occurrence of load level at T hours isThe load data in the study period is expressed as l= { L ₁ ,L ₂ ,···,L _t ,···,L _T }。

3. The method for evaluating reliability of short-term running capacity of wind power based on payload capacity according to claim 1, characterized by: in the step S5, M wind turbines are added into the initial system, and the ACD method is used for calculating the loss of load probability LOLP of the new system _t,M+K The method comprises the following steps:

s51, the effective capacity distribution of the wind turbine j is expressed as:

wherein: j=1, 2, …, M, P _W,j (t) is the output of the wind turbine j at the moment t,for the effective capacity distribution of the wind turbine j, P _FOR,Wj The forced outage probability of the wind turbine j is set;

s52, output P of wind turbine j _W,j For discrete random variables, a discrete random variable P is defined _W,j The v-order moment of (2) is:

p _Wj ＝1-P _FOR,Wj

wherein: p is p _Wj P is the normal operation probability of the wind turbine j _W,j Output of wind turbine j, beta _v V-order moment which is the effective capacity of the wind turbine generator, wherein v is a positive integer;

s53, converting each order moment into each order center moment, wherein the formula is as follows:

wherein: m is M _Wv Is the v-order central moment of the effective capacity of the wind turbine,taking n as v and arranging and combining;

s54, obtaining the accumulation amount of each order through the center distance of each order, wherein the relation between the accumulation amount of the first 8 orders and the center moment of each order is as follows:

wherein: k (K) _Wv V-order cumulant for the effective capacity of the wind turbine;

s55, the accumulated quantity of each step of equivalent effective capacity of M wind turbines and K thermal power turbines is expressed as:

wherein: KW (KW) _v For the v-order cumulant of equivalent effective capacity of M wind turbines and K thermal power turbines, KS _v The v-order cumulant for equivalent effective capacity of K thermal power generating units, K _Wj,v The v-order cumulant of the wind turbine j;

s56, the distribution of equivalent effective capacity is represented by accumulation through edge worth series expansion, and the distribution function of the effective capacity after the loading of M wind turbines and K thermal power turbines is represented by F _M+K (x) The representation is:

wherein: f (F) _M+K (x) Probability that the power generation capacity provided after loading of M wind turbines and K thermal power turbines is smaller than x; n (x) is a standard normal density function; n (N) ^(γ) (x) Gamma derivative for N (x); g _Wv The normalized cumulative quantity is of v order, and sigma is the standard deviation;

s57, load level L for t hour _t Using F _M+K (L _t ) Indicating that the power generation capacity at t hour is less than L _t To obtain the probability of load loss LOLP at t hour _t,M+K And F is equal to _M+K (L _t ) Is related to LOLP _t,M+K ＝F _M+K (L _t )。

4. The method for evaluating reliability of short-term running capacity of wind power based on payload capacity according to claim 1, characterized by: in the step S7, the short-term operation capacity reliability SOCC of the wind power is obtained _t Solving according to the following formula:

wherein: SOCC (solid State control) _t For the reliability of short-term running capacity of the wind power at the t-th hour, C _W For the total capacity of the installation of the wind turbineQuantity, deltaL _t Adding installed total capacity to initial System of C _W The load quantity which can be borne more at the t hour after the wind turbine generator.